Navigation Strategy by Contact Sensing Interaction for a Biped Humanoid Robot

摘要

This report presents a basic contact interaction-based navigation strategy for a biped humanoid robot to support current visual-based navigation. The robot's arms were equipped with force sensors to detect physical contact with objects. We proposed a motion algorithm consisting of searching tasks, self-localization tasks, correction of locomotion direction tasks and obstacle avoidance tasks. Priority was given to right-side direction to navigate the robot locomotion. Analysis of trajectory generation, biped gait pattern, and biped walking characteristics was performed to define an efficient navigation strategy in a biped walking humanoid robot. The proposed algorithm is evaluated in an experiment with a 21-dofs humanoid robot operating in a room with walls and obstacles. The experimental results reveal good robot performance when recognizing objects by touching, grasping, and continuously generating suitable trajectories to correct direction and avoid collisions.